mindie部署DeepSeek-V3.2--Exp-W8A8报错
t('forum.solved') 已解决
发表于2025-11-21 20:00:39
使用2.2.T31-800I-A2-py311-openeuler24.03-lts-arm64 部署 DeepSeek-V3.2-Exp-W8A8时 失败。
启动时似乎报:
  File "/usr/local/Ascend/atb-models/atb_llm/utils/mapping.py", line 532, in get_tp_group 

    module.num_group = self.world_size // module.group_size 

                       ~~~~~~~~~~~~~~~~^^~~~~~~~~~~~~~~~~~~ 

ZeroDivisionError: integer division or modulo by zero

和 [Errno 2] No such file or directory: '//16399_0_execute_request' 的错误,详细如下,请问具体是什么原因,有人用 mindie 2.2.T31 版本部署成功过 DeepSeek-V3.2吗?




[2025-11-21 19:54:14,407] [16446] [281471352762080] [llm] [INFO] [request_router.py-69] : init base config 

[2025-11-21 19:54:14,407] [16446] [281471352762080] [llm] [INFO] [metrics.py-56] : profiling is disenabled. 

[2025-11-21 19:54:28,300] [16437] [281471820099296] [llmmodels] [INFO] [model_runner.py-160] : model_runner.quantize: w8a8, model_runner.kv_quant_type: None, model_runner.fa_quant_type: None, model_runner.dtype: torch.bfloat16 

Traceback (most recent call last): 

  File "/usr/local/lib/python3.11/site-packages/mindie_llm/connector/request_router/request_router.py", line 23, in run 

    self._target(*self._args, **self._kwargs) 

  File "/usr/local/lib/python3.11/site-packages/mindie_llm/connector/request_router/request_router.py", line 99, in do_inference 

    self.initialize(execute_request.config) 

  File "/usr/local/lib/python3.11/site-packages/mindie_llm/connector/request_router/request_router.py", line 84, in initialize 

    initialize_result = self.router_impl.initialize(config) 

                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 

  File "/usr/local/lib/python3.11/site-packages/mindie_llm/connector/request_router/router_impl.py", line 166, in initialize 

    self.generator = Generator(model_config={**model_config.model_config}) 

                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 

  File "/usr/local/lib/python3.11/site-packages/mindie_llm/text_generator/generator.py", line 204, in __init__ 

    self.generator_backend = get_generator_backend(model_config) 

                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 

  File "/usr/local/lib/python3.11/site-packages/mindie_llm/text_generator/adapter/__init__.py", line 32, in get_generator_backend 

    return generator_cls(model_config) 

           ^^^^^^^^^^^^^^^^^^^^^^^^^^^ 

  File "/usr/local/lib/python3.11/site-packages/mindie_llm/text_generator/adapter/generator_torch.py", line 134, in __init__ 

    super().__init__(model_config) 

  File "/usr/local/lib/python3.11/site-packages/mindie_llm/text_generator/adapter/generator_backend.py", line 152, in __init__ 

    self.model_wrapper = get_model_wrapper(model_config, backend_type) 

                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 

  File "/usr/local/lib/python3.11/site-packages/mindie_llm/modeling/model_wrapper/__init__.py", line 15, in get_model_wrapper 

    return wrapper_cls(**model_config) 

           ^^^^^^^^^^^^^^^^^^^^^^^^^^^ 

  File "/usr/local/lib/python3.11/site-packages/mindie_llm/modeling/model_wrapper/atb/atb_model_wrapper.py", line 35, in __init__ 

    self.model_runner = ModelRunner( 

                        ^^^^^^^^^^^^ 

  File "/usr/local/Ascend/atb-models/atb_llm/runner/model_runner.py", line 185, in __init__ 

    self.mapping = Mapping(world_size=world_size, rank=rank, llm_config=self.llm_config, **kwargs) 

                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 

  File "/usr/local/Ascend/atb-models/atb_llm/utils/mapping.py", line 133, in __init__ 

    self.get_tp_group(self.moe_ep_intra_node) 

  File "/usr/local/Ascend/atb-models/atb_llm/utils/mapping.py", line 532, in get_tp_group 

    module.num_group = self.world_size // module.group_size 

                       ~~~~~~~~~~~~~~~~^^~~~~~~~~~~~~~~~~~~ 

ZeroDivisionError: integer division or modulo by zero 

/usr/lib64/python3.11/multiprocessing/resource_tracker.py:254: UserWarning: resource_tracker: There appear to be 6 leaked shared_memory objects to clean up at shutdown 

  warnings.warn('resource_tracker: There appear to be %d ' 

/usr/lib64/python3.11/multiprocessing/resource_tracker.py:267: UserWarning: resource_tracker: '//16399_0_execute_request': [Errno 2] No such file or directory: '//16399_0_execute_request' 

  warnings.warn('resource_tracker: %r: %s' % (name, e)) 

/usr/lib64/python3.11/multiprocessing/resource_tracker.py:267: UserWarning: resource_tracker: '//16399_0_shared_sync_link_response': [Errno 2] No such file or directory: '//16399_0_shared_sync_link_response' 

  warnings.warn('resource_tracker: %r: %s' % (name, e)) 

/usr/lib64/python3.11/multiprocessing/resource_tracker.py:267: UserWarning: resource_tracker: '//16399_0_transfer_response': [Errno 2] No such file or directory: '//16399_0_transfer_response' 

  warnings.warn('resource_tracker: %r: %s' % (name, e)) 

/usr/lib64/python3.11/multiprocessing/resource_tracker.py:267: UserWarning: resource_tracker: '//16399_0_shared_sync_link_request': [Errno 2] No such file or directory: '//16399_0_shared_sync_link_request' 

  warnings.warn('resource_tracker: %r: %s' % (name, e)) 

/usr/lib64/python3.11/multiprocessing/resource_tracker.py:267: UserWarning: resource_tracker: '//16399_0_transfer_request': [Errno 2] No such file or directory: '//16399_0_transfer_request' 

  warnings.warn('resource_tracker: %r: %s' % (name, e)) 

/usr/lib64/python3.11/multiprocessing/resource_tracker.py:267: UserWarning: resource_tracker: '//16399_0_execute_response': [Errno 2] No such file or directory: '//16399_0_execute_response' 

  warnings.warn('resource_tracker: %r: %s' % (name, e)) 

Daemon is killing... 

All subprocesses exited gracefully.

发表于2025/11/21